The Beveridge curve is widely considered a useful tool to study labour markets. It is also referred to as the UV space as it represents the relationship between unemployment (U) and job vacancies (V). The curve provides important insights into the functioning of labour markets by showing changes with regard to two main characteristics: labour market tightness and matching efficiency. Visualising the Beveridge curve does not only show how difficult or easy it is for people to find jobs during different phases of the business cycle. More importantly, the trajectory of the curve gives us clues about structural factors that might influence a labour market's performance in connecting people to jobs.
The main concepts behind the Beveridge curve
The graph below provides a simplified model of the Beveridge curve. The vertical axis depicts the job vacancy rate; the horizontal axis shows the unemployment rate. The job vacancy rate is defined as:number of job vacancies / (number of occupied posts + number of job vacancies) * 100. The graph contains a solid and a dotted curve. It illustrates some basic ideas about how to analyse trends and their potential causes with the help of the Beveridge curve.
Let's focus first on the solid black curve. The curve is decreasing from the upper-left corner of the graph to the lower-right. Movements along the curve's trajectory are typically the result of the regular ups and downs of the business cycle. When the ratio between the vacancy rate to the unemployment rate increases (towards the upper-left), this indicates a better economic situation with higher growth and job availability. In contrast, moving down the curve implies that there are fewer jobs relative to the number of unemployed individuals as a result of lower growth or a recession.
The 45 degree dotted line divides the space into two areas that are associated with expansionary or recessionary trends in the job market. The area below the line is characterised by a tighter labour market than the area above the line as more unemployed individuals compete for fewer open positions.
In reality, the situation of the labour market does not only move up and down the trajectory of a fixed curve. There are other factors that influence how easily people find jobs that cannot be explained by economic growth alone. In particular, the curve can shift towards the inside or the outside, i.e. closer to or away from the origin. This movement is depicted in the graph by the dotted curve that is more distant from the origin than the solid black curve.
A shift towards the outside indicates a deterioration of the labour market's matching efficiency. Matching refers to the ability of the job market to bring together available jobs with individuals looking for a job. Improving the matching process shifts the curve towards the origin of the chart.
In the graph, the solid curve has a better matching efficiency than the dotted curve. Imagine that two economies have the same vacancy rate. If their Beveridge curves are on different trajectories, they will have different levels of unemployment with the same amount of available jobs. The economy with a curve closer to the origin will achieve a lower unemployment rate with the same vacancy rate.
There can be many reasons why the Beveridge curve of an economy might shift towards or away from the origin. At the face of it, a curve shifting outside can mean that the skills in demand are different from the available skills. Usually, there are deeper structural causes why matching improves or worsens, including labour mobility, labour market policies, as well as demographic and technological change.
With the fundamentals of the Beveridge curve in place, let's move on to an example with real data.
Trends in the German labour market from 2010 to 2023
The next chart shows the Beveridge curve for Germany. It uses quarterly data from the fourth quarter of 2010 to the final quarter of 2023. The data for the vacancy rate and the unemployment rate come from Eurostat and are seasonally adjusted. You can find more details about the data in the code repository of my blog.
The visualisation is adapted from the example of a connected scatterplot in the D3.js gallery. The curve gradually builds up from start to finish using an animation that allows to follow the movement across time. Press the replay button to restart the animation.
Germany's Beveridge curve from 2010 to 2023
Data from Eurostat.
The shape of the curve roughly resembles the curves in the diagram from above. The choice of the time period determines the start and direction of the movement along the curve. 2010 was still marked by the consequences of the global financial crisis. Therefore, the curve begins in the lower-right area associated with economic slowdown, although Germany weathered the crisis rather well with lower unemployment rates than in countries hit harder by the crisis. The process of recovery is clearly visible in the quick reduction of unemployment in the following years.
A key observation is that Germany temporarily managed to shift its Beveridge curve closer to the origin. Until 2017, the unemployment rate decreased on average faster than the vacancy rate increased. For instance, in the fourth quarter of 2017, Germany's unemployment rate was about 3 per cent lower than four years before, but the vacancy rate was only approximately half a percentage point higher. In other words, Germany needed less vacancies per new job filled, and the matching efficiency increased.
The curve then creates a small loop on the inner trajectory. First going upwards with a steep rise in the vacancy rate. The Covid pandemic and its economic fallout put a stop to the job boom. However, the decline in vacancies and the increase in unemployment initially remain on the inner trajectory with the higher matching efficiency. For instance, at the lowest point in the inner loop (2020-Q2) Germany has a similar vacancy rate like at the beginning of 2013, but a much lower unemployment rate.
A potential explanation for why Germany initially remained on the more efficient inner loop at the beginning of the Covid pandemic could be found in its active labour market policies. During the Covid crisis, Germany reverted to so-called "Kurzarbeit", a policy already used during the global financial crisis a decade earlier. This policy incentivises employers to keep their staff on the payroll with fewer working days per week and subsidised wages.
The policy ended at the end of 2021. However, Germany's Beveridge curve had already left the inner trajectory earlier than that. During a brief period between the second and third quarters 2020 unemployment rose slightly while vacancy rates increased. As a result, the curve jumped back on its original course.
From mid-2020 until 2023, the vacancy rate rose steeply, while unemployment - already at a very low level - only decreased slightly. During 2023, the vacancy rate dropped, but the unemployment rate remained stable. Overall, the German labour market remained in the upper-left area of the chart despite a challenging international context leading to a spike in energy prices in Germany.
At the end of 2023, Germany's unemployment rate was so low that it can hardly be reduced any further by economic growth alone with open vacancies remaining at a relatively high level. Hence, it is not surprising that the labour market remains a hot topic in German domestic politics. The public debate regularly touches upon all kinds of structural questions, ranging from shortages of skilled labour to the consequences of many elderly people leaving the labour force as a result of an aging society.
Conclusion
There is certainly a lot more to say about the German labour market during this period. A comparison with other countries would also contribute some interesting insights. But I leave it here as a more detailed analysis is outside the scope of this blog post. However, the example shows that the Beveridge curve is a great tool to tell the story of a labour market over time. It also helps to identify turning points and develop questions that deserve deeper analysis.
As always the code for the data collection and visualisation can be found in the repository of my blog.